|Publication number||US5430734 A|
|Application number||US 08/017,519|
|Publication date||Jul 4, 1995|
|Filing date||Feb 12, 1993|
|Priority date||Feb 12, 1993|
|Publication number||017519, 08017519, US 5430734 A, US 5430734A, US-A-5430734, US5430734 A, US5430734A|
|Inventors||Kent L. Gilson|
|Original Assignee||Metalithic Systems, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (14), Non-Patent Citations (2), Referenced by (114), Classifications (18), Legal Events (8)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This patent application is related to my U.S. Pat. No. 5,361,373 entitled "Integrated Circuit Computing Device Comprising a Dynamically Configurable Gate Array Having a Microporocessor and Reconfigurable Instruction Execution Means and Method Therefor", dated Nov. 11, 1994, which is incorporated herein by reference.
1. Field of the Invention
This invention generally relates to integrated circuit devices, and, more specifically, relates to a fault-tolerant integrated circuit device comprising a wafer of dynamically configurable gate arrays. This device provides fault-tolerance with respect to manufacturing defects by mapping all defective gate arrays and defective portions of each gate array on the wafer. This mapping of defects occurs when the wafer is initially tested after fabrication, and the mapping information is used to program the wafer with the desired functions without using any of the defective portions of the wafer.
2. Description of the Related Art
Programmable logic devices are well-known in the electronics art, and have progressed from simple AND-OR arrays to very complex Field Programmable Gate Arrays (FPGAs), which have a large number of input/output (I/O) blocks, programable logic blocks and programmable routing resources to interconnect the logic blocks to each other and to the I/O blocks. Many uses for these FPGAs have been found, with most being used to implement a high number of combinatorial logic functions, which results in lower part count, lower power dissipation, higher speed and greater system flexibility than if discrete components were used.
In recent years FPGAs based on Random Access Memory (RAM) were introduced by several manufacturers, including XILINX. The basic configuration of the XILINX FPGA is described in U.S. Pat. No. 4,870,302 to Freeman, which is assigned to XILINX, and is incorporated herein by reference. In addition, the technical features of XILINX FPGAs are described in XILINX, The Programmable Gate Array Data Book, (1992). The XILINX RAM-based FPGA has multiple I/O blocks, logic blocks and routing resources. The routing resources are used to interconnect the logic blocks to each other and to the I/O blocks, and to connect the I/O blocks to the I/O pads of the FPGA. The programming of the FPGA is accomplished by loading Configuration Data into the Configuration Memory Array of the FPGA. Since the XILINX FPGA is RAM-based, when power is first applied to the FPGA it has not yet been configured. Once the Configuration Data has been loaded into the Configuration Memory Array, the FPGA is ready for operation. The process of designing a circuit for a XILINX FPGA is described in XILINX, User Guide and Tutorials, (1991).
For a brief history of advances in integrated circuit (IC) technology, see John L. Hennessy and David A. Patterson, Computer Architecture: A Quantitative Approach, 53-62, (1990). With each advancement in IC technology, performance has typically increased by one to two orders of magnitude. The driving force behind IC performance includes three basic factors: 1) gate propagation delay, or time it takes a transistor to turn on and off; 2) signal propagation delay, or time it takes a signal to propagate from the output of one gate to the input of another; and 3) level of integration or number of gates that are incorporated onto a working die.
The first two factors, gate and signal propagation delay, are determined by the minimum feature size of the processing technology. As the minimum feature size is reduced, the effects of both of these factors are reduced, resulting in an improvement in overall speed and performance. The third factor, integration density, is determined primarily by the clean room environment and the processing technology. As the level of integration increases, the circuit will become more powerful, since more logical resources can be fabricated on-chip. This higher level of integration eliminates the long propagation delays that occur between two chips, increasing system speed and performance.
The obvious solution to increasing the speed of the IC would be to increase the overall size of the die and decrease the size of each gate or processing element. However, as the die size increases, the number of devices on a semiconductor wafer decreases and the yield (percentage of acceptable working die on one wafer) drops exponentially. This drop is expected since the number of defects per given area remains constant, while the size of the die increases, which increases the likelihood that a defect will occur in the die. Yield is determined by two primary factors, the particle count in parts per billion (PPB) of the clean room, and the die size of the chip. If an error, due to dust or inconsistencies in the crystal lattice, occurs on a specific chip on the wafer, then the chip becomes useless. Since yield drops as chip size increases, chips are typically designed with only the minimum circuits required to make them work, and if any part of the chip is non-functioning then the entire chip is useless. As the chip size increases, the chip count on a wafer decreases. As the chip count on a wafer decreases, the probability of chip failure increases. Semiconductor devices with relatively small die size provide a much higher number of devices per wafer and have a much higher yield than for devices that have larger die sizes. Thus, practical limitations on die size directly conflict with the goal of increasing the complexity and functionality of FPGAs, which inherently pushes the die size for FPGAs to increase.
A powerful example of the yield decrease that occurs from increasing die size follows. On a six inch wafer, 23.2 chips that are one cm per side (which are smaller than the 80486 microprocessor developed by Intel) can be built with a typical yield of 3.6%. This yield means that two wafers at a cost of $550 each would be required to get one usable chip. If the die size of the chip were decreased, more chips could be fabricated on a single wafer, and the probability of chip failure would decrease.
Defects in the wafer would have a less dramatic impact if extra or spare circuitry were included on the die to allow bypassing the defect. The manufacturer of the wafer would then be able to determine defects on the die and replace nonfunctional circuitry with the spare circuitry. This technique makes the chip fault-tolerant, and has been implemented by manufacturers to some extent. For example, U.S. Pat. No. 4,937,475 to Rhodes et al. discloses an integrated circuit chip with multiple logic blocks interconnected by a series of horizontal and vertical conductors. The vertical conductors are electrically isolated from the crossing horizontal conductors, and have a laser diffusible region at each intersection such that a laser beam can cause the region to conduct thereby connecting the vertical conductor with the crossing horizontal conductor. In addition the laser beam is used to sever the conductors in certain places. In this manner the laser beam can custom-configure the chip as desired by making some connections and breaking others, which allows the defective portions of the chip to be bypassed.
The success of fault-tolerance through redundant or spare circuitry has been limited, however, because the granularity of the spare circuitry in current architectures is unworkable. For example, in the case of a microprocessor chip, to duplicate functional units (such as the Arithmetic Logic Unit (ALU), registers, etc.), the designer would have to increase the area of the die by at least twice. If nothing was wrong with the primary units, then half of the die would be wasted on unused spares. In addition, increasing the size of the die to accommodate spares causes the yield to drop even further, since the same yield statistics that apply to an entire chip also apply to the spares.
A solution to this problem is to provide a general-purpose functional block architecture similar to the XILINX dynamically configurable FPGAs referenced above. If each functional block on a chip is identical, faults on the chip can be easily bypassed by routing the circuitry around any functional blocks rendered non-functional by defects. If the granularity of the functional blocks is sufficiently small, the number of defective functional blocks will be small compared to the number of operable functional blocks. Thus an FPGA which has functional blocks of sufficiently small granularity will not be significantly impacted by defects, making the FPGA fault-tolerant. A wafer of such FPGAs would allow bypassing of all defects, rendering the entire wafer fault-tolerant. In this manner one large circuit could be made using the entire area of a wafer. This process, called waferscale integration, would allow circuits to be developed and fabricated that are fault-tolerant and, hence, not susceptible to the traditional drop in yield associated with increasing the size of the die.
Therefore, there existed a need to provide a waferscale integrated circuit device and method which provides a high level of complexity and functionality while still maintaining high production yields through a programmable functional block architecture and a fault-tolerance scheme that utilizes the functional blocks of the device that are functional after fabrication and does not utilize the functional blocks of the device that are non-functional after fabrication. These functional blocks must be of sufficiently small granularity to assure a high yield of functional blocks with respect to manufacturing defects.
It is an object of this invention to provide a fault-tolerant integrated circuit device and method comprising a wafer of configurable gate arrays which are interconnected such that non-functioning gate arrays on the wafer are not used.
It is another object of this invention to provide a fault-tolerant integrated circuit device and method comprising a wafer of XILINX FPGAs which are interconnected such that non-functioning FPGAs on the wafer are not used.
It is a further object of this invention to provide a fault-tolerant integrated circuit computing device and method comprising a wafer of dynamically configurable gate arrays programmed to implement a microprocessor and a reconfigurable instruction execution unit which can be dynamically configured to implement a variety of high level functions in hardware, thereby drastically improving system speed.
According to the first embodiment of the present invention, a fault-tolerant integrated circuit device is provided. This device is a wafer of FPGAs such as the RAM-based XC3020 FPGA by XILINX. Once fabrication of the wafer is complete, each FPGA on the wafer is functionally tested. All FPGAs rendered non-functional due to defects are mapped as non-functional, and also any portions of FPGAs that are non-functional without rendering the entire FPGA non-functional are likewise mapped. This defect map is used to assure the defective portions of the wafer are not used. The I/O pads of the gate arrays on the wafer are then interconnected with bonding wires. The interconnection scheme with the bonding wires may interconnect all FPGAs on the wafer, but nominally the FPGAs that are completely non-functional are bypassed and not connected to the functional FPGAs.
The desired function of the wafer may be represented in a number of different forms. A high level Functional Description could include either computer code comprising a list of instructions of the desired function, a list of equations relating the outputs of the circuit to functions of the inputs and other outputs, a list of formulae, or other high-level functional descriptions. This high level Functional Description is then compiled using special purpose software into a Hardware Description of the desired circuit functions. In the alternative, the desired Hardware Description could be entered directly via a schematic diagram, or could also be represented using a Virtual Hardware Description Language (VHDL), which is well-known in the art. The Hardware Description is a general description of the desired circuit functions that is independent of the specific implementation of the circuitry on the wafer.
Another special software program compiles the Hardware Description of the desired wafer function into a list of required XILINX resources, which comprise a subset of the available resources on the wafer, which comprise the I/O blocks, logic blocks and routing resources within the FPGAs on the wafer. This is possible due to the functional block architecture of the XILINX FPGAs, which provides a finite set of circuit functions and configurations that each functional block can provide. A Partitioner then partitions the resultant set of required XILINX resources among the many FPGAs on the wafer using the map of defects to assure that only functional portions of the wafer are used. This partitioning defines the function of each individual FPGA on the wafer. A Mapper then uses the defect map to determine the interconnects required within each FPGA without using any defective portions of the FPGA, and creates the Configuration Data for each individual FPGA.
When the wafer is initially powered up, it is then configured by loading the appropriate Configuration Data into the Configuration Memory Array of each FPGA. Since the XILINX FPGAs are RAM-based, they can be dynamically reconfigured to implement any number of desired functions. But the configuration data for each new configuration of the wafer must go through the same partitioning and mapping process to assure that no defective portions of the wafer are used.
In the second embodiment of the present invention, a wafer of XILINX FPGAs is not used, but the device of the present invention is implemented using a custom wafer which nominally has an array of RAM-based, dynamically configurable FPGAs, along with other circuitry specially suited to specific applications.
In the third embodiment of the present invention, one or more of the FPGAs on the wafer have a microprocessor and reconfigurable instruction execution unit as described in my related patent application. The number of logic blocks and routing resources must allow for implementing a simple microprocessor such as a Reduced Instruction Set Computer (RISC). This RISC Processor is coupled to a Reconfigurable Instruction Execution Unit that can be reconfigured to implement a variety of very complex operations in hardware. This is accomplished by dynamically reprogramming the FPGA such that the Reconfigurable Instruction Execution Unit is reconfigured to perform the appropriate operation in hardware while most of the remaining functions within the FPGA remain unaltered. This configuration allows the reconfigurable instruction execution unit to implement very complex operations relatively quickly in hardware rather than using time-consuming software routines, thereby greatly improving system speed. If the wafer has 100 functional FPGAs which are all interconnected with each programmed with a RISC Processor and Reconfigurable Instruction Execution Unit, this array of computing devices could share tasks and operate in parallel, which would greatly improve the computational speed for a given task. The wafer would then be a computer with 100 different computing devices all interconnected on the same wafer, allowing for a high level of parallel processing and, hence, a very high level of performance and computing speed not available with small computers of the prior art.
The foregoing and other objects, features and advantages will be apparent from the following description of the preferred embodiment of the invention as illustrated in the accompanying drawings.
FIG. 1 is a top view of the integrated circuit device of the present invention comprising a wafer that has many individual Field Programmable Gate Arrays (FPGAs).
FIG. 2A is a cross-sectional view of the wafer in FIG. 1 taken along the line 2A--2A before the bonding wires are placed to interconnect the FPGAs.
FIG. 2B is a cross-sectional view of the wafer of FIG. 1 taken along the line 2A--2A after the bonding wires are placed to interconnect the FPGAs.
FIG. 2C is a cross-sectional view of the wafer of FIG. 1 taken along the line 2C--2C after the bonding wires are placed to interconnect the FPGAs in a manner to avoid using a non-functional FPGA.
FIG. 3 is a block diagram of one of the many FPGAs fabricated on the wafer shown in FIG. 1.
FIG. 4 is a block diagram of one of the many FPGAs of FIG. 1 when used as a computing device in a high-speed computer system.
FIG. 5 is a flow chart of the Fabrication/Test, Program Generation, and Programming processes used to prepare the device 10 of FIG. 1 for operation.
An understanding of the basic design and programming process for XILINX FPGAs will aid the reader in understanding the device of the present invention. A reference which explains in detail the design flow and methodology for XILINX FPGAs is XILINX, User Guide and Tutorials, (1991).
The fault-tolerant integrated circuit device 10 of the present invention is shown in FIG. 1. This device 10 comprises a wafer 12 which has many Field Programmable Gate Arrays (FPGAs) 14 fabricated thereon. While only the top half of wafer 12 is shown fully populated with FPGAs 14, this is done for illustrative purposes and it is evident that the entire wafer 12 is used in reality. The detailed configuration of each individual FPGA 14 is shown in FIG. 3, and includes I/O pads 30, I/O Blocks 32, Logic Blocks 34, Routing Resources (not shown) and a Configuration Memory Array 60. The I/O Pads 30 are contacts on the FPGA 14 which connect the FPGA 14 to circuitry external to FPGA 14. The I/O Blocks 32 are programmable blocks that can be configured to provide input or output signals to the I/O Pads 30 of FPGA 14, and can also be programmed with such features as signal feedback into FPGA 14 or registered inputs and outputs.
The Logic Blocks 34 contain circuitry that can be programmed to perform a myriad of different functions. The specific configuration of the Logic Blocks 34 is not critical, provided the Logic Blocks 34 provide the required minimum level of functionality to implement the desired circuitry. The Routing Resources (not shown) are a large number of conductors that cross at numerous points within FPGA 14 controlled by programmable pass transistors, and, once properly programmed, the pass transistors and Routing Resources provide the majority of the signal paths between the I/O Pads 30, the I/O Blocks 32, and the Logic Blocks 34. These programmable pass transistors are programmed by writing Configuration Data to Configuration Memory Array 60. The XILINX XC3020 FPGA can be programmed by writing Configuration Data to Configuration Memory Array 60 in either parallel or serial modes. In either case, the Configuration Memory Array 60 is filled with Configuration Data from an external source, which programs the FPGA 14 to perform the desired functions.
The FPGA 14 is thus a general purpose, programmable device. The I/O Blocks 32 and the Logic Blocks 34 can each be programmed to perform specific functions, and the Routing Resources can be programmed to interconnect the I/O Blocks 32 with the I/O Pads 30 and the Logic Blocks 34 to achieve the desired overall function of the FPGA 14. The result of this flexible architecture is a user-configurable FPGA capable of a very high level of performance.
FIGS. 2A-2C illustrate the interconnection of FPGAs 14 on wafer 12 using bonding wires. FIG. 2A shows a cross-sectional view of two FPGAs 14A and 14B prior to interconnecting these with bonding wires. FIG. 2B shows these same FPGAs once bonding wires 11 are in place to connect some of the I/O pads of FPGA 14A to some of the I/O pads on FPGA 14B. FIG. 2C illustrates how a non-functional FPGA 14D can be bypassed by connecting FPGA 14C directly to FPGA 14E using bonding wires 11.
FIG. 5 shows the flow chart for preparing the device 10 of the present invention for operation, and the first portion of interest is the Fabrication/Test process. The first step is to fabricate the wafer. After fabrication is complete, each FPGA 14 on wafer 12 is functionally tested, and a Defect Map is created. This Defect Map can be stored in a special, non-volatile memory portion 16 of the wafer 12 as shown in FIG. 1, or can also be stored on a floppy diskette. The Defect Map contains two different types of data: 1) Location of FPGAs which are completely non-functional, and 2) Location of non-functional portions of partially functional FPGAs. This completes the Fabrication/Test portion of the process for generating the device 10 as shown in FIG. 5.
The person desiring to use the device 10 must develop a program for configuring the device 10 appropriately to accomplish the desired task. This portion of the design process is represented in the Program Generation process in FIG. 5. The first step is to create a Hardware Description of the desired circuit functions. This can be done directly using a schematic editor or a Virtual Hardware Description Language (VHDL) to directly represent the desired circuit functions. In the alternative, a Functional Description in the form of high-level computer code, equations or formulae could be compiled into an equivalent Hardware Description. Since the Functional Description can be precluded by using a direct Hardware Description, the Functional Description block in FIG. 5 is represented using dashed lines to show that this step is optional.
The Hardware Description is used as input to a software design tool which takes the Hardware Description and compiles it into a list of required XILINX resources to accomplish the desired function. The available XILINX resources depend on the specific number and type of functional blocks provided on the particular XILINX FPGA being used, and the number of FPGAs 14 on the wafer 12. A Partitioner then takes the list of required XILINX resources and allocates the list of required XILINX resources among the individual FPGAs 14 on the wafer 12, being sure not to allocate any required resources to any FPGAs 14 that are substantially non-functional according to the Defect Map. A standard XILINX Router then places and routes the required resources for each individual FPGA, avoiding non-functional portions according to the Defect Map. A standard XILINX Bit-stream Generator is then used to generate Configuration Data for configuring each FPGA appropriately. This step is repeated until Configuration Data for each FPGA 14 on wafer 12 is created. This completes the Program Generation Process shown in FIG. 5.
Once the Configuration Data for the entire wafer 12 is created, the Programming process shown in FIG. 5 loads the Configuration Data into the appropriate FPGAs 14, and the device 10 is ready for operation. Since the XILINX FPGAs are RAM-based, it is possible to dynamically reconfigure one or all of the FPGAs on wafer 12 to change the function of device 10 as needed. However, all Configuration Data for wafer 12 must be generated using the Program Generation process of FIG. 5 to assure the defects on wafer 12 are bypassed.
Since the Configuration Data for each FPGA 14 depends on the Defect Map, which is specific to a given wafer, the Configuration Data for one wafer would not work with a second wafer, since it may attempt to use resources on the second wafer that are non-functional. Thus the Configuration Data is unique for each wafer, depending on the number and location of defects. This feature can help eliminate software piracy, since the software (which includes Configuration Data) for one wafer would not work on a second wafer.
According to the second embodiment of the device 10 of the present invention, a custom wafer is used rather than a wafer of XILINX FPGAs. This custom wafer will have many functional blocks that are essentially configurable FPGAs, but may also include special-purpose circuitry to optimize the device for a particular application. In this custom wafer application, different portions of the wafer could be directly interconnected using a polysilicon or metal layer on the wafer rather than using bonding wires. If all functional blocks on the entire wafer are interconnected, there must be some means for breaking the connection to non-functional blocks. A laser beam could easily etch around all the non-functional blocks, thereby breaking their connection to the functional portions of the wafer. Even with a custom wafer, the same wire bonding process used in the first embodiment may be used to interconnect the functional blocks on the custom wafer, allowing the bonding wires to bypass non-functional portions of the wafer.
According to the third embodiment of the present invention, one or more of the XILINX FPGAs in the first embodiment are programmed to be a Computing Device 50, as shown in FIG. 4. The Computing Device 50 is implemented within FPGA 14, and has a RISC Processor 54, which is coupled to a Reconfigurable Instruction Execution Unit 56, a Host Interface (I/F) 58, and a Configuration Memory Array 60. The Host Interface 58 is coupled to an external Host 40 via System Bus 44. The Host 40 controls the configuration of FPGA 14 by loading new Configuration Data through Host Interface 58 into the Configuration Memory Array 60 of FPGA 14 as required by the specific operation. RISC Processor 54 is coupled to Program Memory 42 which contains the code (instructions) for RISC Processor 54.
Many previous applications of FPGAs include very complex combinatorial circuits, or various sequencers or state machines. The most common use of the FPGA is to replace a large number of discrete integrated circuits by implementing these many functions in a single FPGA. But as the FPGA has increased in complexity and gate count, the complexity of the functions that could be implemented in the FPGA has also dramatically increased. With the Reduced Instruction Set Computer (RISC) architecture that requires so much less circuitry than the traditional Complex Instruction Set Computer (CISC) microprocessor, it is now possible to implement a small RISC processor within a portion of the XILINX FPGA 14. But the implementation of a RISC processor within an FPGA is not, in and of itself, greatly significant. Any person skilled in the microprocessor art could implement a RISC processor within an FPGA given the necessary level of complexity and functionality within the FPGA.
The Computing Device 50 within FPGA 14, however, is not a typical RISC processor. Included within FPGA 14 is a Reconfigurable Instruction Execution Unit 56. A typical RISC processor has a fixed instruction execution unit where all data manipulations are performed. The Reconfigurable Execution Unit 56 of the present invention allows the FPGA 14 to execute extremely complex instructions in a very short time, which greatly boosts the speed and performance of the Computing Device 50.
The operation of the Computing Device 50 is best understood by referring to FIG. 4. Once the FPGA 14 is powered up, it must be initially configured or programmed by the Host 40. The Host 40 thus writes the appropriate Configuration Data into the Configuration Memory Array 60, which programs the FPGA 14 to the appropriate initial state. The RISC processor 54 then begins executing its program from Program Memory 42. For illustrative purposes with the system shown in FIG. 4, it is assumed that the Host 40 is the "master" and the Computing Device 50 within FPGA 14 is a "slave" to Host 40. In this configuration the Host 40 is a computer capable of general-purpose functions. As the Host 40 executes its program, it may encounter a complex, time-consuming operation. When a complex operation is needed that would require a relatively long time to execute, and which can be implemented instead in hardware, Host 40 reconfigures the FPGA 14 for that particular operation by initiating the programming sequence for the FPGA 14 which causes new Configuration Data to be written into the Configuration Memory Array 60. Now referring to FIGS. 3 and 4, the effect of this new Configuration Data is to leave the programming of the I/O Blocks 32 unchanged, to leave the programming of the Logic Blocks 34 that comprise the RISC processor 54 unchanged, but to change the programming of the Logic Blocks 34 that comprise the Reconfigurable Instruction Execution Unit 56 such that the desired complex operation can be accomplished by the newly configured hardware on data that already exists within the Reconfigurable Instruction Execution Unit 56. Once the desired operation is complete, the FPGA 14 can signal the Host 40 and transfer the data to Host 40, if required.
This operation allows the Computing Device 50 within FPGA 14 to execute many different and complex operations in hardware rather than in long, time-consuming software routines. Since the Reconfigurable Instruction Execution Unit 56 within Computing Device 50 can be dynamically reconfigured to accomplish each of these complex operations in programmable hardware rather than employing fixed, special-purpose circuitry for each operation, the Computing Device 50 provides great flexibility and power with a minimum of circuitry.
Since the Computing Device 50 described above can be implemented in one XILINX FPGA, and the wafer 12 has many XILINX FPGAs, it is clear that multiple Computing Devices 50 may be implemented on a single wafer 12. These Computing Devices 50 are all controlled by a common Host 40 via a common System Bus 44, and each nominally have their own Program Memory 42. In this manner a single wafer can be programmed to provide a large number of Computing Devices 50 which can all operate in parallel, sharing tasks at the control of the Host 40. This allows for a waferscale Computing Device with fault-tolerance and performance that rivals and even surpasses some of the most powerful supercomputers known in the prior art.
It is an obvious modification that the Computing Device 50 of FIG. 4 could also be implemented on a custom wafer such as the custom wafer of the second embodiment rather than on a wafer of XILINX FPGAs. Other modifications of this embodiment are also obvious. For example, the RISC Processor 54 could be the "master" with the Host 40 being a "slave" which simply reconfigures FPGA 14 on command. In addition, the FPGA 14 need not necessarily be RAM-based if developments in other technologies such as EEPROMS make the use of other types of reconfigurable circuitry desirable. While the disclosure herein describes only the Reconfigurable Instruction Execution Unit 56 as changing during reconfiguration of the FPGA 14, it is an obvious extension of this invention that the attributes of the RISC Processor 54, the Host Interface 58, and other circuitry within the FPGA 14 could all be modified as needed during reconfiguration of the FPGA 14.
While the invention has been described in its preferred embodiment, it is to be understood that the words which have been used are words of description rather than limitation, and that changes may be made within the purview of the appended claims without departing from the true scope and spirit of the invention in its broader aspects. For example, while the discussion herein specifically refers to RAM-based, dynamically configurable gate arrays, it is obvious that gate arrays that are only configured once after fabrication (statically configurable, not dynamically configurable) could also be made fault-tolerant with the same approach taught herein.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4140967 *||Jun 24, 1977||Feb 20, 1979||International Business Machines Corporation||Merged array PLA device, circuit, fabrication method and testing technique|
|US4380811 *||Apr 14, 1981||Apr 19, 1983||International Business Machines Corp.||Programmable logic array with self correction of faults|
|US4418410 *||Dec 8, 1980||Nov 29, 1983||International Business Machines Corporation||Error detection and correction apparatus for a logic array|
|US4511812 *||Jun 15, 1982||Apr 16, 1985||Hitachi, Ltd.||Programmable logic array, including an arrangement for invalidating faulty and term outputs|
|US4811345 *||Dec 16, 1986||Mar 7, 1989||Advanced Micro Devices, Inc.||Methods and apparatus for providing a user oriented microprocessor test interface for a complex, single chip, general purpose central processing unit|
|US4857773 *||Aug 1, 1988||Aug 15, 1989||Ricoh Co., Ltd.||Programming logic device with test-signal enabled output|
|US4870302 *||Feb 19, 1988||Sep 26, 1989||Xilinx, Inc.||Configurable electrical circuit having configurable logic elements and configurable interconnects|
|US4937475 *||Sep 19, 1988||Jun 26, 1990||Massachusetts Institute Of Technology||Laser programmable integrated circuit|
|US5109353 *||Dec 2, 1988||Apr 28, 1992||Quickturn Systems, Incorporated||Apparatus for emulation of electronic hardware system|
|US5128559 *||Aug 8, 1991||Jul 7, 1992||Sgs-Thomson Microelectronics, Inc.||Logic block for programmable logic devices|
|US5212652 *||Aug 15, 1989||May 18, 1993||Advanced Micro Devices, Inc.||Programmable gate array with improved interconnect structure|
|US5315535 *||Jun 25, 1993||May 24, 1994||Nec Corporation||Automatic router capable of searching for a new wiring with reference to connection failures|
|US5321322 *||Mar 24, 1993||Jun 14, 1994||Aptix Corporation||Programmable interconnect architecture without active devices|
|US5341092 *||Oct 7, 1992||Aug 23, 1994||Actel Corporation||Testability architecture and techniques for programmable interconnect architecture|
|1||*||Babb, Tessier and Agarwal, Virtual Wires: Overcoming Pin Limitations in FPGA based Logic Emulators Jan. 26, 1993 pp. 1 15.|
|2||Babb, Tessier and Agarwal, Virtual Wires: Overcoming Pin Limitations in FPGA-based Logic Emulators Jan. 26, 1993 pp. 1-15.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US5537631 *||Mar 21, 1994||Jul 16, 1996||Current Logic Systems, Inc.||Memory management technique for defective memory|
|US5539752 *||Jun 30, 1995||Jul 23, 1996||Advanced Micro Devices, Inc.||Method and system for automated analysis of semiconductor defect data|
|US5737631 *||Apr 5, 1995||Apr 7, 1998||Xilinx Inc||Reprogrammable instruction set accelerator|
|US5748979 *||Jun 7, 1995||May 5, 1998||Xilinx Inc||Reprogrammable instruction set accelerator using a plurality of programmable execution units and an instruction page table|
|US5777887 *||May 12, 1995||Jul 7, 1998||Crosspoint Solutions, Inc.||FPGA redundancy|
|US5794062 *||Apr 17, 1995||Aug 11, 1998||Ricoh Company Ltd.||System and method for dynamically reconfigurable computing using a processing unit having changeable internal hardware organization|
|US5815404 *||Oct 16, 1995||Sep 29, 1998||Xilinx, Inc.||Method and apparatus for obtaining and using antifuse testing information to increase programmable device yield|
|US5818726 *||Mar 6, 1996||Oct 6, 1998||Cadence Design Systems, Inc.||System and method for determining acceptable logic cell locations and generating a legal location structure|
|US5991907 *||Oct 11, 1996||Nov 23, 1999||Lucent Technologies Inc.||Method for testing field programmable gate arrays|
|US6003150 *||Nov 20, 1997||Dec 14, 1999||Lucent Technologies Inc.||Method for testing field programmable gate arrays|
|US6058469 *||May 11, 1998||May 2, 2000||Ricoh Corporation||System and method for dynamically reconfigurable computing using a processing unit having changeable internal hardware organization|
|US6070004 *||Sep 25, 1997||May 30, 2000||Siemens Aktiengesellschaft||Method of maximizing chip yield for semiconductor wafers|
|US6108806 *||Apr 13, 1998||Aug 22, 2000||Lucent Technologies Inc.||Method of testing and diagnosing field programmable gate arrays|
|US6167558 *||Feb 20, 1998||Dec 26, 2000||Xilinx, Inc.||Method for tolerating defective logic blocks in programmable logic devices|
|US6191613 *||Jul 29, 1999||Feb 20, 2001||Xilinx, Inc.||Programmable logic device with delay-locked loop|
|US6202182||Jun 30, 1998||Mar 13, 2001||Lucent Technologies Inc.||Method and apparatus for testing field programmable gate arrays|
|US6256758||Mar 3, 1999||Jul 3, 2001||Agere Systems Guardian Corp.||Fault tolerant operation of field programmable gate arrays|
|US6530049||Jul 6, 2000||Mar 4, 2003||Lattice Semiconductor Corporation||On-line fault tolerant operation via incremental reconfiguration of field programmable gate arrays|
|US6550030||Sep 27, 1999||Apr 15, 2003||Lattice Semiconductor Corp.||On-line testing of the programmable logic blocks in field programmable gate arrays|
|US6574761||Sep 27, 1999||Jun 3, 2003||Lattice Semiconductor Corp.||On-line testing of the programmable interconnect network in field programmable gate arrays|
|US6631487||Sep 27, 2000||Oct 7, 2003||Lattice Semiconductor Corp.||On-line testing of field programmable gate array resources|
|US6735755 *||Mar 27, 2001||May 11, 2004||Jeng-Jye Shau||Cost saving methods using pre-defined integrated circuit modules|
|US6803785 *||Jun 12, 2001||Oct 12, 2004||Altera Corporation||I/O circuitry shared between processor and programmable logic portions of an integrated circuit|
|US6915212||May 8, 2003||Jul 5, 2005||Moac, Llc||Systems and methods for processing complex data sets|
|US6931580||Mar 13, 2000||Aug 16, 2005||International Business Machines Corporation||Rapid fail analysis of embedded objects|
|US6980024||Sep 8, 2004||Dec 27, 2005||Altera Corporation||I/O circuitry shared between processor and programmable logic portions of an integrated circuit|
|US6996470||Oct 15, 2003||Feb 7, 2006||Moac Llc||Systems and methods for geophysical imaging using amorphous computational processing|
|US7127697 *||Jul 30, 2003||Oct 24, 2006||Xilinx, Inc.||Methods of utilizing programmable logic devices having localized defects in application-specific products|
|US7148505 *||May 13, 2005||Dec 12, 2006||Actel Corporation||Method for using a chip carrier substrate with a land grid array and external bond terminals|
|US7228521 *||Feb 1, 2005||Jun 5, 2007||Xilinx Inc.||System for operating a programmable logic device having a defective region|
|US7251804 *||Oct 1, 2004||Jul 31, 2007||Xilinx, Inc.||Structures and methods of overcoming localized defects in programmable integrated circuits by routing during the programming thereof|
|US7284229 *||Oct 1, 2004||Oct 16, 2007||Xilinx, Inc.||Multiple bitstreams enabling the use of partially defective programmable integrated circuits while avoiding localized defects therein|
|US7308608 *||May 1, 2002||Dec 11, 2007||Cypress Semiconductor Corporation||Reconfigurable testing system and method|
|US7310758 *||Aug 22, 2005||Dec 18, 2007||Xilinx, Inc.||Circuit for and method of implementing programmable logic devices|
|US7394708||Mar 18, 2005||Jul 1, 2008||Xilinx, Inc.||Adjustable global tap voltage to improve memory cell yield|
|US7412635 *||Oct 1, 2004||Aug 12, 2008||Xilinx, Inc.||Utilizing multiple bitstreams to avoid localized defects in partially defective programmable integrated circuits|
|US7424655 *||Oct 1, 2004||Sep 9, 2008||Xilinx, Inc.||Utilizing multiple test bitstreams to avoid localized defects in partially defective programmable integrated circuits|
|US7446561||Nov 17, 2005||Nov 4, 2008||Altera Corporation||I/O circuitry shared between processor and programmable logic portions of an integrated circuit|
|US7478261||Sep 13, 2005||Jan 13, 2009||M2000||Reconfigurable circuit with redundant reconfigurable cluster(s)|
|US7493472||Jul 29, 2005||Feb 17, 2009||Ricoh Company Ltd.||Meta-address architecture for parallel, dynamically reconfigurable computing|
|US7619438||Oct 11, 2007||Nov 17, 2009||Xilinx, Inc.||Methods of enabling the use of a defective programmable device|
|US7620863||Jul 29, 2008||Nov 17, 2009||Xilinx, Inc.||Utilizing multiple test bitstreams to avoid localized defects in partially defective programmable integrated circuits|
|US7644385 *||Nov 7, 2005||Jan 5, 2010||Altera Corporation||Programmable logic device with performance variation compensation|
|US7737724||Dec 27, 2007||Jun 15, 2010||Cypress Semiconductor Corporation||Universal digital block interconnection and channel routing|
|US7757198||Oct 19, 2007||Jul 13, 2010||Lattice Semiconductor Corporation||Scan chain systems and methods for programmable logic devices|
|US7761690||Jul 26, 2007||Jul 20, 2010||International Business Machines Corporation||Method, apparatus and computer program product for dynamically selecting compiled instructions|
|US7761845||Sep 9, 2002||Jul 20, 2010||Cypress Semiconductor Corporation||Method for parameterizing a user module|
|US7765095||Nov 1, 2001||Jul 27, 2010||Cypress Semiconductor Corporation||Conditional branching in an in-circuit emulation system|
|US7770113||Nov 19, 2001||Aug 3, 2010||Cypress Semiconductor Corporation||System and method for dynamically generating a configuration datasheet|
|US7774190||Nov 19, 2001||Aug 10, 2010||Cypress Semiconductor Corporation||Sleep and stall in an in-circuit emulation system|
|US7797651 *||Dec 11, 2007||Sep 14, 2010||Xilinx, Inc.||Verifying design isolation using bitstreams|
|US7810059||Oct 11, 2007||Oct 5, 2010||Xilinx, Inc.||Methods of enabling the validation of an integrated circuit adapted to receive one of a plurality of configuration bitstreams|
|US7825688||Apr 30, 2007||Nov 2, 2010||Cypress Semiconductor Corporation||Programmable microcontroller architecture(mixed analog/digital)|
|US7844437||Nov 19, 2001||Nov 30, 2010||Cypress Semiconductor Corporation||System and method for performing next placements and pruning of disallowed placements for programming an integrated circuit|
|US7849435||Jul 29, 2008||Dec 7, 2010||Xilinx, Inc.||Utilizing multiple test bitstreams to avoid localized defects in partially defective programmable integrated circuits|
|US7853916||Oct 11, 2007||Dec 14, 2010||Xilinx, Inc.||Methods of using one of a plurality of configuration bitstreams for an integrated circuit|
|US7865862||Nov 8, 2007||Jan 4, 2011||International Business Machines Corporation||Design structure for dynamically selecting compiled instructions|
|US7893724||Nov 13, 2007||Feb 22, 2011||Cypress Semiconductor Corporation||Method and circuit for rapid alignment of signals|
|US8000826 *||Jan 24, 2006||Aug 16, 2011||Synopsys, Inc.||Predicting IC manufacturing yield by considering both systematic and random intra-die process variations|
|US8010826||Nov 11, 2009||Aug 30, 2011||Meta Systems||Reconfigurable circuit with redundant reconfigurable cluster(s)|
|US8026739||Dec 27, 2007||Sep 27, 2011||Cypress Semiconductor Corporation||System level interconnect with programmable switching|
|US8040266||Mar 31, 2008||Oct 18, 2011||Cypress Semiconductor Corporation||Programmable sigma-delta analog-to-digital converter|
|US8049569||Sep 5, 2007||Nov 1, 2011||Cypress Semiconductor Corporation||Circuit and method for improving the accuracy of a crystal-less oscillator having dual-frequency modes|
|US8067948||Feb 21, 2007||Nov 29, 2011||Cypress Semiconductor Corporation||Input/output multiplexer bus|
|US8069405||Nov 19, 2001||Nov 29, 2011||Cypress Semiconductor Corporation||User interface for efficiently browsing an electronic document using data-driven tabs|
|US8069428||Jun 12, 2007||Nov 29, 2011||Cypress Semiconductor Corporation||Techniques for generating microcontroller configuration information|
|US8069436||Aug 10, 2005||Nov 29, 2011||Cypress Semiconductor Corporation||Providing hardware independence to automate code generation of processing device firmware|
|US8078894||Mar 27, 2008||Dec 13, 2011||Cypress Semiconductor Corporation||Power management architecture, method and configuration system|
|US8078970||Nov 9, 2001||Dec 13, 2011||Cypress Semiconductor Corporation||Graphical user interface with user-selectable list-box|
|US8085067||Dec 21, 2006||Dec 27, 2011||Cypress Semiconductor Corporation||Differential-to-single ended signal converter circuit and method|
|US8085100||Feb 19, 2008||Dec 27, 2011||Cypress Semiconductor Corporation||Poly-phase frequency synthesis oscillator|
|US8092083||Oct 1, 2007||Jan 10, 2012||Cypress Semiconductor Corporation||Temperature sensor with digital bandgap|
|US8103496||Nov 1, 2001||Jan 24, 2012||Cypress Semicondutor Corporation||Breakpoint control in an in-circuit emulation system|
|US8103497||Mar 28, 2002||Jan 24, 2012||Cypress Semiconductor Corporation||External interface for event architecture|
|US8117580||Aug 23, 2007||Feb 14, 2012||Xilinx, Inc.||Multiple bitstreams enabling the use of partially defective programmable integrated circuits while avoiding localized defects therein|
|US8120408||Jul 14, 2008||Feb 21, 2012||Cypress Semiconductor Corporation||Voltage controlled oscillator delay cell and method|
|US8130025||Apr 17, 2008||Mar 6, 2012||Cypress Semiconductor Corporation||Numerical band gap|
|US8149048||Aug 29, 2001||Apr 3, 2012||Cypress Semiconductor Corporation||Apparatus and method for programmable power management in a programmable analog circuit block|
|US8160864||Nov 1, 2001||Apr 17, 2012||Cypress Semiconductor Corporation||In-circuit emulator and pod synchronized boot|
|US8176296||Oct 22, 2001||May 8, 2012||Cypress Semiconductor Corporation||Programmable microcontroller architecture|
|US8358150||Oct 11, 2010||Jan 22, 2013||Cypress Semiconductor Corporation||Programmable microcontroller architecture(mixed analog/digital)|
|US8370791||Jun 3, 2008||Feb 5, 2013||Cypress Semiconductor Corporation||System and method for performing next placements and pruning of disallowed placements for programming an integrated circuit|
|US8402313||Nov 20, 2007||Mar 19, 2013||Cypress Semiconductor Corporation||Reconfigurable testing system and method|
|US8476928||Aug 3, 2011||Jul 2, 2013||Cypress Semiconductor Corporation||System level interconnect with programmable switching|
|US8499270||Jun 28, 2011||Jul 30, 2013||Cypress Semiconductor Corporation||Configuration of programmable IC design elements|
|US8516025||Apr 16, 2008||Aug 20, 2013||Cypress Semiconductor Corporation||Clock driven dynamic datapath chaining|
|US8533677||Sep 27, 2002||Sep 10, 2013||Cypress Semiconductor Corporation||Graphical user interface for dynamically reconfiguring a programmable device|
|US8555032||Jun 27, 2011||Oct 8, 2013||Cypress Semiconductor Corporation||Microcontroller programmable system on a chip with programmable interconnect|
|US8595670 *||Mar 8, 2010||Nov 26, 2013||Altera Corporation||Method and apparatus for circuit block reconfiguration EDA|
|US8717042||Nov 29, 2011||May 6, 2014||Cypress Semiconductor Corporation||Input/output multiplexer bus|
|US8736303||Dec 16, 2011||May 27, 2014||Cypress Semiconductor Corporation||PSOC architecture|
|US8793635||Nov 28, 2011||Jul 29, 2014||Cypress Semiconductor Corporation||Techniques for generating microcontroller configuration information|
|US8796740 *||Jan 13, 2009||Aug 5, 2014||Altera Corporation||Using a single mask for various design configurations|
|US8909960||Jul 8, 2011||Dec 9, 2014||Cypress Semiconductor Corporation||Power management architecture, method and configuration system|
|US9147611||Jun 17, 2014||Sep 29, 2015||Altera Corporation||Using a single mask for various design configurations|
|US9448964||Apr 22, 2010||Sep 20, 2016||Cypress Semiconductor Corporation||Autonomous control in a programmable system|
|US9564902||Dec 31, 2007||Feb 7, 2017||Cypress Semiconductor Corporation||Dynamically configurable and re-configurable data path|
|US9720805||Mar 28, 2008||Aug 1, 2017||Cypress Semiconductor Corporation||System and method for controlling a target device|
|US20040225443 *||May 8, 2003||Nov 11, 2004||Moac Llc||Systems and methods for processing complex data sets|
|US20050027455 *||Oct 15, 2003||Feb 3, 2005||Moac Llc||Systems and methods for geophysical imaging using amorphous computational processing|
|US20050268070 *||Jul 29, 2005||Dec 1, 2005||Baxter Michael A||Meta-address architecture for parallel, dynamically reconfigurable computing|
|US20060186917 *||Nov 17, 2005||Aug 24, 2006||Altera Corporation, A Corporation Of Delaware||I/O circuitry shared between processor and programmable logic portions of an integrated circuit|
|US20060259432 *||Jul 17, 2006||Nov 16, 2006||Sony Corporation||Information processing apparatus, information processing method, and providing medium|
|US20070057693 *||Sep 13, 2005||Mar 15, 2007||M2000||Reconfigurable circuit with redundant reconfigurable cluster(S)|
|US20070174797 *||Jan 24, 2006||Jul 26, 2007||Jianfeng Luo||Predicting IC manufacturing yield by considering both systematic and random intra-die process variations|
|US20070228548 *||Jun 12, 2007||Oct 4, 2007||Actel Corporation||Chip carrier substrate with a land grid array and external bond terminals|
|US20080259998 *||Oct 1, 2007||Oct 23, 2008||Cypress Semiconductor Corp.||Temperature sensor with digital bandgap|
|US20080297388 *||Mar 31, 2008||Dec 4, 2008||Cypress Semiconductor Corporation||Programmable sigma-delta analog-to-digital converter|
|US20090031111 *||Jul 26, 2007||Jan 29, 2009||Chou Deanna J||Method, apparatus and computer program product for dynamically selecting compiled instructions|
|US20090066427 *||Feb 19, 2008||Mar 12, 2009||Aaron Brennan||Poly-phase frequency synthesis oscillator|
|US20090125704 *||Nov 8, 2007||May 14, 2009||International Business Machines Corporation||Design structure for dynamically selecting compiled instructions|
|US20100095147 *||Nov 11, 2009||Apr 15, 2010||Abound Logic, S.A.S.||Reconfigurable circuit with redundant reconfigurable cluster(s)|
|WO1996036140A1 *||May 10, 1996||Nov 14, 1996||Crosspoint Solutions, Inc.||Fpga redundancy|
|WO2007031249A1 *||Sep 11, 2006||Mar 22, 2007||M2000 Sa||Reconfigurable circuit with redundant reconfigurablf cluster(s)|
|U.S. Classification||714/725, 714/E11.111, 326/38, 714/724, 257/E21.526, 711/101, 711/170, 716/117|
|International Classification||H01L21/66, G06F11/20, G11C29/00|
|Cooperative Classification||H01L2924/0002, H01L22/22, G06F17/5054, G11C29/006|
|European Classification||G06F17/50D4, H01L22/22, G11C29/00W|
|Feb 12, 1993||AS||Assignment|
Owner name: NATIONAL TECHNOLOGY, INC., LOUISIANA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST.;ASSIGNOR:GILSON, KENT L.;REEL/FRAME:006439/0661
Effective date: 19930210
|Mar 20, 1995||AS||Assignment|
Owner name: METALITHIC SYSTEMS, INC.
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:NATIONAL TECHNOLOGY, INC.;REEL/FRAME:007403/0157
Effective date: 19940829
|Jan 9, 1998||AS||Assignment|
Owner name: WAGNER & BROWN, LTD., TEXAS
Free format text: SECURITY AGREEMENT;ASSIGNOR:METALITHIC SYSTEMS, INC.;REEL/FRAME:008876/0231
Effective date: 19980107
|Jan 5, 1999||AS||Assignment|
Owner name: WAGNER & BROWN, LTD., TEXAS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:METALITHIC SYSTEMS, INC.;REEL/FRAME:009670/0977
Effective date: 19981207
|Jan 12, 1999||AS||Assignment|
Owner name: SEASOUND, LLC, CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:WAGNER & BROWN, LTD.;REEL/FRAME:009689/0441
Effective date: 19981207
|Jan 26, 1999||REMI||Maintenance fee reminder mailed|
|Jul 4, 1999||LAPS||Lapse for failure to pay maintenance fees|
|Sep 14, 1999||FP||Expired due to failure to pay maintenance fee|
Effective date: 19990704